Artificial Neural Network Based Image Detection Applied to Foot Massage Machines

Abstract

For nearly two decades, many massage machine (MM) manufacturers have developed a lot of distinct types of FMM. Common massage goods on the sale are roller and pressing models. However, stimulating all acupuncture points (AP) is extremely hard for distinct sizes of feet accurately. Besides, the massage roller cannot be manipulated all alone. Thence, the author proposed a novel computer vision skillfulness to make out the foot acupuncture points (FAP) by ANN. First, the sole of users’ soles is captured and image preprocessing procedures are executed to segment the region of interest (ROI) of soles. FAP is mapped to foot images (FI) to obtain reference massage positions. Afterwards, the YCbCr color space is used to part the brightness to get done the segmentation of the FI in the skin detection. To improve the success rate of image segmentation (IS), ANN is used to train plantar image set. Finally, a FMM was redesigned to raise the rate of ID and user convenience. Experimental results confirm the practicality of the proposed ID method for FMM.